Creating a Geospatial and Visual Information Ontology for Analysts

An ontology is a main component of an evolving knowledge base that caters to multiple clients. Consider a scenario where an automated procedure (a computer vision algorithm) used in an analyst tool detects different kinds of “roads” in images, and features in the ontology are used to distinguish a “paved” road from a “dirt road”. In another scenario, the ontology enables reasoning about “locations”, supporting analysts' geospatial information processing tasks. In this paper, we describe the creation of a multiuse geospatial and visual information ontology, GVIO 1 , building on and integrating with the lexical database, WordNet. To ensure that GVIO can interoperate with other ontologies in useful ways, we inherit as much of the WordNet structure and content as is relevant for the domain of aerial surveillance and link in new content/structure as necessary.